【16.50%】【CF 44G】Shooting Gallery

time limit per test

5 seconds

memory limit per test

256 megabytes

input

standard input

output

standard output

Berland amusement park shooting gallery is rightly acknowledged as one of the best in the world. Every day the country's best shooters master their skills there and the many visitors compete in clay pigeon shooting to win decent prizes. And the head of the park has recently decided to make an online version of the shooting gallery. During the elaboration process it turned out that the program that imitates the process of shooting effectively, is needed. To formulate the requirements to the program, the shooting gallery was formally described. A 3D Cartesian system of coordinates was introduced, where the X axis ran across the gallery floor along the line, along which the shooters are located, the Y axis ran vertically along the gallery wall and the positive direction of the Z axis matched the shooting direction. Let's call the XOY plane a shooting plane and let's assume that all the bullets are out of the muzzles at the points of this area and fly parallel to the Z axis. Every clay pigeon can be represented as a rectangle whose sides are parallel to X and Y axes, and it has a positive z-coordinate. The distance between a clay pigeon and the shooting plane is always different for every target. The bullet hits the target if it goes through the inner area or border of the rectangle corresponding to it. When the bullet hits the target, the target falls down vertically into the crawl-space of the shooting gallery and cannot be shot at any more. The targets are tough enough, that's why a bullet can not pierce a target all the way through and if a bullet hits a target it can't fly on. In input the simulator program is given the arrangement of all the targets and also of all the shots in the order of their appearance. The program should determine which target was hit by which shot. If you haven't guessed it yet, you are the one who is to write such a program.

Input

The first line contains an integer n (1 ≤ n ≤ 105) — the number of targets. Each of the subsequent n lines contains the description of a target. The target is described by five integers xl, xr, yl, yr, z, that determine it's location in space (0 ≤ xl < xr ≤ 107, 0 ≤ yl < yr ≤ 107, 0 < z ≤ 107). The next line contains an integer m (1 ≤ m ≤ 105), determining the number of shots. Then in m lines shots are described. Every shot is determined by the coordinates of a bullet on the shooting plane (x, y) (0 ≤ x, y ≤ 107, the coordinates of bullets are integers). The shots are given in the order of their firing. The intervals between shots are large enough, and a target falls very quickly, that's why assume that a falling target can not be an obstruction for all the shots following the one that hit it.

Output

For every shot in the single line print the number of the target which the shot has hit, or 0, if the bullet did not hit any target. The targets are numbered starting from 1 in the order in which they were given in the input data.

Examples

input

2
1 4 1 4 1
2 5 2 6 2
4
0 0
3 3
4 5
3 5

output

0
1
2
0

【题解】

意思是说每个靶子离射击点的距离为z.然后描述矩形靶子的左下角和右下角。

射击m次。每次射击靶子之后。靶子会毁坏(假如同一个点有多个靶子重叠，最前面那个靶子坏掉。其他靶子还可以射击)；

做法:

给每个子弹记录三个信息x,y,bianhao。即坐标以及各个子弹出现的先后顺序。

然后以靶子离射击点的距离为关键字从小到大排序。

处理第一个靶子是被哪一个子弹射击了。即在靶子所在范围内的子弹。且要求子弹的编号(射击顺序)最小。

这个问题可以用kd-tree来解决。

以子弹的点为元素建立kdtree

知道一个靶子是被哪个子弹射击之后。在kdtree中删掉这个子弹。然后继续处理下一个靶子。

编程复杂度很高。

【代码】

#include <cstdio>
#include <algorithm>

using namespace std;

const int MAXN = 105000;
const int INF = 2100000000;

struct target
{
	int mi_n[2], ma_x[2], z, n;
};

struct point
{
	int min, n, dot, d[2],fa,l,r;
};

int n, m, root, now, txl, txr, tyl, tyr, ans[MAXN] = { 0 };

target rec[MAXN];
point t[MAXN];

void input_data()
{
	scanf("%d", &n);
	for (int i = 1; i <= n; i++)
	{
		scanf("%d%d%d%d%d", &rec[i].mi_n[0], &rec[i].ma_x[0], &rec[i].mi_n[1], &rec[i].ma_x[1], &rec[i].z);
		rec[i].n = i;
	}
	scanf("%d", &m);
	for (int i = 1; i <= m; i++)
	{
		scanf("%d%d", &t[i].d[0], &t[i].d[1]);
		t[i].n = i;
	}
}

bool cmp_1(point a, point b)
{
	return a.d[now] < b.d[now];
}

void gengxin(int father, int son)
{
	if (t[father].min > t[son].min)
	{
		t[father].min = t[son].min;
		t[father].dot = t[son].dot;
	}
}

void up_data(int rt)
{
	t[rt].min = t[rt].n; t[rt].dot = rt; //dot可以说是当前这个子树的编号最小的点的节点。
	if (t[rt].n == 0) //如果点已经删掉了
	{
		t[rt].min = INF;
		t[rt].dot = 0;
	}
	int l = t[rt].l, r = t[rt].r;
	if (l)
		gengxin(rt, l);
	if (r)
		gengxin(rt, r);
}

int build(int begin, int end, int fa,int fx)
{
	int m = (begin + end) >> 1;
	now = fx;
	nth_element(t + begin, t + m, t + end + 1, cmp_1);
	t[m].fa = fa;
	if (begin < m)
		t[m].l = build(begin, m - 1, m, 1 - fx);
	if (m < end)
		t[m].r = build(m + 1, end, m, 1 - fx);
	up_data(m);
	return m;
}

bool inrange(int x, target b,int fx)
{
	return ((b.mi_n[fx] <= x) && (x <= b.ma_x[fx]));
}

int query(int rt, int fx, int r, int xl, int xr, int yl, int yr)
{//r是当前处理的靶子编号 txl,txr,tyl,tyr是靶子信息。
	//xl,xr,yl,yr是当前kd-tree的节点所在子树的所有点的范围(如果不确定就写成无穷大)
	if (!rt)//我们会不断缩小这个范围。
		return 0;
	if (!t[rt].dot)
		return 0;
	if (txl <= xl && xr <= txr && tyl <= yl && yr <= tyr)
		return t[rt].dot;
	int t1 = 0,t2 = 0;
	if (inrange(t[rt].d[fx], rec[r], fx))//这个节点的fx坐标在靶子范围内
	{ //则左右儿子都要递归求解。则后序还要查看当前这个节点是否能更新。
		if (fx == 0) //如果是以x轴排序
		{
			t1 = query(t[rt].l, 1-fx, r, xl, t[rt].d[0], yl, yr);
			t2 = query(t[rt].r, 1-fx, r, t[rt].d[0], xr, yl, yr);
		}
		else//以y轴排序
		{
			t1 = query(t[rt].l, 1-fx, r, xl, xr, yl, t[rt].d[1]);
			t2 = query(t[rt].r, 1-fx, r, xl, xr, t[rt].d[1], yr);
		}
	}
	else//这个靶子的xy坐标没有完全把这个点包住
	{//则只要返回左或右儿子
		if (t[rt].d[fx] < rec[r].mi_n[fx])//大于这个节点的fx坐标 等号的情况上面已经考虑了
		{
			if (fx == 0)
				return query(t[rt].r, 1-fx, r, t[rt].d[0], xr, yl, yr);
			else
				return query(t[rt].r, 1-fx, r, xl, xr, t[rt].d[1], yr);
		}
		if (rec[r].ma_x[fx] < t[rt].d[fx])//小于这个节点的fx坐标
		{
			if (fx == 0)
				return query(t[rt].l, 1-fx, r, xl, t[rt].d[0], yl, yr);
			else
				return query(t[rt].l, 1-fx, r, xl, xr, yl, t[rt].d[1]);
		}
	}
	int fi = 0; int temp = INF;
	if (t1!=0)
		if (t[t1].n < temp)//千万不要写成t[t1].min < temp
		{//因为可能t[t1].min对应的点不在我们所需的范围内 
			//这个rt就已经返回的是所需的节点的编号了,不要乱来
			temp = t[t1].n;
			fi = t1;
		}
	if (t2 != 0)
		if (t[t2].n < temp)
		{
			temp = t[t2].n;
			fi = t2;
		}
	if (t[rt].n != 0 && t[rt].n < temp)//这是尝试用当前这个节点来更新。
		if (inrange(t[rt].d[0],rec[r],0) && inrange(t[rt].d[1],rec[r],1))
			{
				temp = t[rt].n;
				fi = rt;
			}
	return fi;
}

void adjust(int rt) //删掉一个点后调整相关点的信息
{
	up_data(rt);
	if (rt != root)
		adjust(t[rt].fa);
}

bool cmp_2(target a, target b)
{
	return a.z < b.z;
}

void get_ans()
{
	root = build(1, m, 0, 0);
	sort(rec + 1, rec + 1 + n, cmp_2);
	for (int i = 1; i <= n; i++)
	{
		txl = rec[i].mi_n[0], txr = rec[i].ma_x[0], tyl = rec[i].mi_n[1], tyr = rec[i].ma_x[1];
		int hit = query(root, 0, i, 0, INF, 0, INF);
		if (hit != 0)
		{
			ans[t[hit].n] = rec[i].n;
			t[hit].n = 0;
			adjust(hit);
		}
	}
}

void output_ans()
{
	for (int i = 1; i <= m; i++)
		printf("%d
", ans[i]);
}

int main()
{
	//freopen("F:\rush.txt", "r", stdin);
	input_data();
	get_ans();
	output_ans();
	return 0;
}